Home
Who We Serve
About Contact Setup a Demo
AI Agents

AI agents, purpose-built for private markets.

CapHive's AI agents work inside the platform to streamline and supercharge ongoing tasks — from onboarding and portfolio analysis to compliance checklists and querying structured and unstructured data.

KYC Onboarding Agentconnected to your fund data
Is Meridian Family Office's KYC ready to rely on?
Reviewed. The record is complete, internally consistent and nothing has expired — so it is cleared before your team relies on it for capital calls, distributions or AML clearance. Onboarding · KYC review
Cash Flow Assistant Agentconnected to your fund data
What is Fund III's cash position and runway?
Calculated, along with the recommended capital call, and the consolidated report has gone to fund managers. These are rules-based calculations — no AI involved in the numbers. Finance · Cash position & runway
Data Quality Agentconnected to your fund data
Anything wrong in our fund data this morning?
Last night's pass covered every fund. It found commitments that have collected more than was committed, valuations that have gone stale and FX rates that have not refreshed — health score and fix-it list already emailed. Data · Nightly integrity checks
Deal Management Agentconnected to your fund data
How healthy is the Series C deal?
Scoped to that one deal: stagnant investors, overdue activity steps, missing data-room documents, extracted terms and per-investor note summaries — finishing with a pipeline digest email. Deals · Single-deal health
Portfolio Report Agentconnected to your fund data
Draft the market section of this company's evaluation.
Generated from the uploaded pitch materials, one section at a time. There is no batch run here — each section is generated or refined on demand. Portfolio · AI portfolio report
The agents

16 agents, each with one job.

These are not one assistant with sixteen names. Each has its own scope, its own permissions and its own record of what it did — and each says plainly whether it uses a model or is just checking your records. Open one for the summary, or read its full guide.

Onboarding

Getting an investor from first document to a record you can rely on.

Onboarding · Agent

E-Sign Monitor

The E-Sign Monitor keeps watch over every document sent out for e-signature within your entity.

Each time it runs it flags signature requests that have been unsigned too long, and documents that expired, were cancelled, voided or failed, then emails a status report to your team.

  • Runs against — The whole entity, in one pass
  • Uses AI — No — rule checks only
  • Output — A saved report, always "ready", plus one combined email
Read the full guide
Onboarding · Agent

KYC Doc Extractor Agent

The KYC Doc Extractor Agent is the document reader behind CapHive's document-first KYC experience: an investor uploads their…

identity and address documents, the agent reads them with AI and pre-fills the matching KYC form fields. The investor reviews and confirms the details rather than typing everything in by hand.

  • Runs against — One investor's KYC record, at upload time
  • Uses AI — Yes — one request per uploaded document
  • Output — Pre-filled KYC fields and an on-screen summary; no stored report
Read the full guide
Onboarding · Agent

KYC Onboarding Agent

The KYC Onboarding Agent reviews a completed investor KYC record and checks that it is complete, internally consistent and not…

expired — before your team relies on it for capital calls, distributions or AML clearance. Operations and compliance teams use it as the final quality check on investor onboarding.

  • Runs against — One investor's KYC record, or every completed record in the entity
  • Uses AI — Yes — for two of the document checks
  • Output — A report on the record, marked Ready or Blocked, plus AML screening and expiry email
Read the full guide
Onboarding · Agent

Side Letter Obligations Agent

Side letters create commitments your team must honour — quarterly reports for one investor, fee arrangements for another…

co-investment rights for a third. This agent reads the side letter attached to each commitment, pulls out every obligation with the exact clause it came from, and lines them all up for your legal team to review.

  • Runs against — One commitment, or every commitment in a fund
  • Uses AI — Yes — for reading obligations out of the document
  • Output — Obligations in the Obligation Matrix, then tracked tasks and alerts once approved
Read the full guide
Finance

Gating capital events and keeping the cash picture current.

Finance · Agent

Call Notice Agent

The Call Notice Agent checks that a capital call is actually ready before its notices reach investors.

It works on one call at a time, walks a fixed checklist — KYC, document generation, document approval and notification setup — and shows live progress on the capital call's own page.

  • Runs against — One capital call, one check at a time
  • Uses AI — No — every check is a yes/no comparison against your existing records
  • Output — A readiness panel on the capital call, plus a saved report marked Ready or Blocked
Read the full guide
Finance · Agent

Cash Flow Assistant Agent

The Cash Flow Assistant Agent is a rules-based reporting tool that checks how up to date your fund data is, calculates each fund's…

cash position, runway and recommended capital call, and emails a consolidated report to fund managers. A secondary chat feature lets users ask questions about the latest results in plain language.

  • Runs against — Every fund listed in its setup, entity-wide
  • Uses AI — No — the reporting pipeline is rules-based and never calls an AI model
  • Output — A per-fund report, a report file in each fund's Reports folder, and one consolidated email
Read the full guide
Finance · Agent

Distribution Readiness Agent

The Distribution Readiness Agent is a rule-based, step-by-step agent that checks every prerequisite is in place — investor…

KYC, computed payment amounts, generated and approved notices, and bank details — before a capital distribution payment goes out. Fund operations teams run it on a distribution immediately before wiring funds.

  • Runs against — One capital distribution, or every distribution in the entity in batch mode
  • Uses AI — No — entirely rule-based, and it works with no AI key configured
  • Output — A live step panel on the distribution, plus a report marked Ready or Blocked
Read the full guide
Finance · Agent

RTA Reconciliation Agent

The RTA Reconciliation Agent reconciles an uploaded registrar and transfer agent unit statement PDF against your fund's own unit…

records in CapHive. It checks for mismatches at scheme (ISIN) and folio level, flags folios in the statement with no matching record, and flags unit classes CapHive holds that the statement never mentions.

  • Runs against — One fund per statement upload
  • Uses AI — Yes — AI reads the statement PDF, so an AI provider key is required
  • Output — A reconciliation report with a health score, plus an optional email summary
Read the full guide
Portfolio

Reading company material and turning it into comparable numbers.

Portfolio · Agent

Investor Presentation Analyst

The Investor Presentation Analyst reads a portfolio company’s investor presentations across reporting periods, extracts…

structured intelligence from each one, and compares periods to surface commitment misses and dropped risks. It produces a pre-board briefing plus a Management Accountability Score, all scoped to one portfolio company at a time.

  • Runs against — One portfolio company at a time
  • Uses AI — Yes — every meaningful step except the score
  • Output — 8–12 grounded pre-board questions and a 0–100 accountability score
Read the full guide
Portfolio · Agent

Portfolio Company Agent

The Portfolio Company Agent keeps an eye on the data health of every portfolio company in your fund — whether KPI reports…

have come in on time, whether required documents are on file, and whether a recent valuation has been recorded. It does not use AI to make judgment calls; it checks your data against the rules you have configured.

  • Runs against — Every portfolio company at once, or a single company
  • Uses AI — No — every check is rule-based
  • Output — A per-company report, plus a portfolio compliance dashboard
Read the full guide
Portfolio · Agent

Portfolio Report Agent

The Portfolio Report Agent is the AI engine behind AI Portfolio Reports — investment-grade evaluation documents generated per…

portfolio company from uploaded pitch materials. It does not behave like a typical support agent: it has no report history of its own, no Schedule tab and no fund- or entity-level batch run, and instead generates or refines one report section at a time, on demand.

  • Runs against — One section of one report
  • Uses AI — Yes — your entity’s configured model, plus OpenAI for refinement
  • Output — A collated report from reviewed sections, exportable to PDF or Word
Read the full guide
Portfolio · Agent

Portfolio Scoring Agent

The Portfolio Scoring Agent scores each portfolio company using a sector-specific scoring guide and AI.

It gathers the company’s documents, KPI data, valuations and investment history, sends all of that to your configured AI model together with the matching scoring guide, and stores a structured 0–100 investment score on the company.

  • Runs against — Every portfolio company with a sector set, or a single company
  • Uses AI — Yes — this agent’s own model setting, or your entity default
  • Output — A new scoring report per run: score, grade, dimensions, data coverage
Read the full guide
Deals

Watching open deals and reporting the state of a raise.

Deals · Agent

Deal Management Agent

The Deal Management Agent monitors the health of a single open deal: stagnant investors, overdue activity steps, missing data-room…

documents, extracted deal terms, per-investor note summaries and AI scores. It runs scoped to one deal at a time — on demand from the deal page or on a weekly schedule — and finishes with a pipeline digest email.

  • Runs against — One open deal at a time
  • Uses AI — Yes — for term extraction, note summaries and scoring
  • Output — Deal report, per-investor reports, pipeline digest email
Read the full guide
Deals · Agent

Deals Report Agent

The Deals Report Agent looks across a fund's open deals and produces a fundraising status report for each one — committed against…

target amounts, fill rate, activity-step distribution, top investors and staleness. It then emails a consolidated summary to whoever you have configured to receive it.

  • Runs against — Every open deal in the organisation, unless narrowed
  • Uses AI — No
  • Output — Per-deal report, an HTML report document, one summary email
Read the full guide
Data

Catching bad data and summarising what people wrote down.

Data · Agent

Data Quality Agent

The Data Quality Agent runs nightly integrity checks across every fund's financial data — commitments that have collected more than…

was committed, valuations that have gone stale, and FX rates that have not been refreshed. It then emails a health score and a fix-it list to the people you have configured to receive it.

  • Runs against — Every fund in the entity, once a day
  • Uses AI — No
  • Output — A health score and fix-it list per fund, by email and in the app
Read the full guide
Data · Agent

Note Summarizer Agent

The Note Summarizer Agent generates an AI-written summary of an investor's notes over the past month or quarter.

It runs across every investor in your entity, can auto-tag untagged notes first, and writes the summary back as a regular note on the investor record so it sits alongside everything else in that investor's notes feed.

  • Runs against — Every investor in the entity, or one investor on demand
  • Uses AI — Yes — for auto-tagging and for the summary itself
  • Output — A summary note on the investor record, plus a report per investor
Read the full guide
How the agents work

Inside your ledger, not beside it.

An agent here is not a general chatbot with your documents attached. It reads and writes against the same records your team uses, under the same permissions, with the same audit trail.

Grounded in your own data

Every answer comes from your funds, investors, portfolio companies and deal documents — not from general knowledge about private markets.

Proposed, permissioned, logged

An agent proposes an action, inherits the permissions of the person who asked, and writes what it did to the audit log.

A human approves the consequential

Anything that moves money or goes to an LP waits for a named person to approve it. The agent prepares; it does not release.

Answers cite the record

Each reply carries the underlying record — the capital event, the statement, the company update or the document clause it came from.

Governance & data handling

Controls before capability.

An agent with access to fund data is a control question, not only a product question. These are the constraints it runs under.

Role-based permissions

An agent sees exactly what the person invoking it is entitled to see — no fund, investor or deal outside that scope.

No training on your data

Your fund, investor and portfolio data is not used to train models. It is used to answer your questions and nothing else.

Full audit log

Every prompt, retrieval and action an agent takes is recorded against a user, a timestamp and a record.

Human-in-the-loop approvals

Capital movement, investor communication and anything filed externally require an explicit human approval step.

Data residency options

Hosting region can be selected so fund data stays in the jurisdiction your structure and LPs require.

Security & Trust

The full control set

Encryption, access control, hosting, retention and the certifications behind the platform.

Questions

What teams ask first.

Where do the agents get their data?

From your CapHive tenant. Fund books, capital events, investor records, portfolio company updates and the documents held in deal rooms and investor files.

Where you have connected banks, custodians or other systems through our integrations, the agents read the records those feeds create — not the external systems directly.

Can an agent send an LP notice on its own?

No. An agent can draft the notice, compute each investor's figures and assemble the distribution list. Release requires a named person to approve it.

The same rule applies to anything that moves money or is filed with a regulator.

What happens when it is unsure?

It says so, and shows what it could not resolve — a missing document, an unaudited number, a figure that two records disagree on.

An agent that cannot ground an answer in a record returns the gap rather than a guess.

Is our data used to train models?

No. Your data is used to answer your questions. It is not used to train models, and it is not pooled with other clients' data.

Our data processing terms set this out in writing. See Security & Trust for the wider control set.

Which structured and unstructured sources can it query?

Structured: fund and investor ledgers, capital accounts, capital events, fee and waterfall computations, portfolio metrics and deal records.

Unstructured: subscription packs, side letters, KYC files, company update decks and reports, and the documents held in a deal room — read within the permissions of the person asking.

Keep going

Where the agents do the work.

See CapHive on your own fund data.

Book a walkthrough with our team — funds, portfolios, deals & secondaries, end to end.

or write to hello@CapHive.com